wbn2007 PT PT I

Point-to-Point Wireless Communication (I):
Digital Basics, Modulation, Detection, Pulse

Shaping
Shiv Kalyanaraman
Google: “Shiv RPI”
shivkuma@ecse.rpi.edu
Based upon slides of Sorour Falahati, A. Goldsmith,

& textbooks by Bernard Sklar & A. Goldsmith
Rensselaer Polytechnic Institute Shivkumar Kalyanaraman
1 : “shiv rpi”
The Basics

2 : “shiv rpi”
Big Picture: Detection under AWGN

3 : “shiv rpi”
Additive White Gaussian Noise (AWGN)
Thermal noise is described by a zero-mean Gaussian random process,
n(t) that ADDS on to the signal => “additive”
Its PSD is flat, hence, it is called white noise.
Autocorrelation is a spike at 0: uncorrelated at any non-zero lag
[w/Hz]
Power spectral
Density
(flat => “white”)
Autocorrelation
Function
Probability density function (uncorrelated)
(gaussian) Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
4 : “shiv rpi”
Effect of Noise in Signal Space
The cloud falls off exponentially (gaussian).
Vector viewpoint can be used in signal space, with a random noise vector w

5 : “shiv rpi”
Maximum Likelihood (ML) Detection: Scalar Case
“likelihoods”
Assuming both symbols equally likely: uA is chosen if
Log-Likelihood => A simple distance criterion!

6 : “shiv rpi”
AWGN Detection for Binary PAM
pz (z | m2 ) pz (z | m1 )
s2 s1
ψ 1 (t )
− Eb 0 Eb
⎛ s1 − s 2 / 2 ⎞
Pe (m1 ) = Pe (m2 ) = Q⎜ ⎟
⎜ N /2 ⎟
⎝ 0 ⎠
⎛ 2 Eb ⎞
PB = PE (2) = Q⎜⎜ ⎟
⎟
⎝ N0 ⎠
7 : “shiv rpi”
Bigger Picture
General structure of a communication systems
Noise
Transmitted Received Received
Info. signal signal info.
SOURCE
Source Transmitter Channel Receiver User
Transmitter
Source Channel
Formatter Modulator
encoder encoder
Receiver
Source Channel
Formatter Demodulator
decoder decoder
8 : “shiv rpi”
Digital vs Analog Comm: Basics

9 : “shiv rpi”
Digital versus analog
Advantages of digital communications:
Regenerator receiver
Original Regenerated
pulse pulse
Propagation distance
Different kinds of digital signal are treated

identically.
Voice
Data A bit is a bit!
Media
10 : “shiv rpi”
Signal transmission through linear systems
Input Output
Linear system
Deterministic signals:
Random signals:
Ideal distortion less transmission:

All the frequency components of the signal not only arrive with
an identical time delay, but also are amplified or attenuated
equally.

11 : “shiv rpi”
Signal transmission (cont’d)
Ideal filters:
Non-causal!
Low-pass
Band-pass Duality => similar problems occur w/

High-pass
rectangular pulses in time domain.
Realizable filters:
RC filters Butterworth filter

12 : “shiv rpi”
Bandwidth of signal
Baseband versus bandpass:
Baseband Bandpass
signal signal
Local oscillator
Bandwidth dilemma:
Bandlimited signals are not realizable!
Realizable signals have infinite bandwidth!
13 : “shiv rpi”
Bandwidth of signal: Approximations
Different definition of bandwidth:
a) Half-power bandwidth d) Fractional power containment bandwidth
b) Noise equivalent bandwidth e) Bounded power spectral density
c) Null-to-null bandwidth f) Absolute bandwidth
(a)
(b)
(c)
(d)
(e)50dB
14 : “shiv rpi”
Formatting and transmission of baseband signal
Digital info.
Textual Format
source info.
Pulse
Analog Transmit
Sample Quantize Encode modulate
info.
Pulse
Bit stream waveforms Channel
Format
Analog
info. Low-pass
Decode Demodulate/
filter Receive
Textual Detect
sink
info.
Digital info.

15 : “shiv rpi”
Sampling of Analog Signals
Time domain Frequency domain
xs (t ) = xδ (t ) × x(t ) X s ( f ) = Xδ ( f ) ∗ X ( f )
x(t )
| X(f )|
xδ (t ) | Xδ ( f ) |
xs (t )
| Xs( f )|

16 : “shiv rpi”
Aliasing effect & Nyquist Rate
LP filter
Nyquist rate
aliasing

17 : “shiv rpi”
Undersampling &Aliasing in
Time Domain

18 : “shiv rpi”
Nyquist Sampling & Reconstruction:
Time Domain
Note: correct reconstruction does not draw straight lines between samples.
Key: use sinc() pulses for reconstruction/interpolation
19 : “shiv rpi”
Nyquist Reconstruction: Frequency Domain
The impulse response of the

reconstruction filter has a classic 'sin(x)/x
shape.
The stimulus fed to this filter is the series

of discrete impulses which are the
samples.

20 : “shiv rpi”
Sampling theorem
Analog Sampling Pulse amplitude

signal process modulated (PAM) signal
Sampling theorem: A bandlimited signal with no spectral

components beyond , can be uniquely determined by
values sampled at uniform intervals of
The sampling rate,
is called Nyquist rate.
In practice need to sample faster than this because the receiving

filter will not be sharp. Shivkumar Kalyanaraman
21 : “shiv rpi”
Quantization
Amplitude quantizing: Mapping samples of a continuous amplitude
waveform to a finite set of amplitudes.
Out
In
Average quantization noise power
Quantized
Signal peak power

values
Signal power to average

quantization noise power

22 : “shiv rpi”
Encoding (PCM)
A uniform linear quantizer is called Pulse Code Modulation

(PCM).
Pulse code modulation (PCM): Encoding the quantized signals
into a digital word (PCM word or codeword).
Each quantized sample is digitally encoded into an l bits
codeword where L in the number of quantization levels and

23 : “shiv rpi”
Quantization error
Quantizing error: The difference between the input and output of a
quantizer
e(t ) = xˆ (t ) − x(t )
Process of quantizing noise

Qauntizer
Model of quantizing noise
y = q( x)
AGC x(t ) xˆ (t )
x(t ) xˆ (t )
x
e(t )
+
e(t ) =
xˆ (t ) − x(t )
24 : “shiv rpi”
Non-uniform quantization
It is done by uniformly quantizing the “compressed” signal.
At the receiver, an inverse compression characteristic, called “expansion” is
employed to avoid signal distortion.
compression+expansion companding
y = C ( x) x̂
x(t ) y (t ) yˆ (t ) xˆ (t )
x ŷ
Compress Qauntize Expand
Transmitter Channel Receiver
25 : “shiv rpi”
Baseband transmission
To transmit information thru physical channels, PCM
sequences (codewords) are transformed to pulses (waveforms).
Each waveform carries a symbol from a set of size M.
Each transmit symbol represents k =log2 M bits of the PCM words.
PCM waveforms (line codes) are used for binary symbols (M=2).
M-ary pulse modulation are used for non-binary symbols

(M>2). Eg: M-ary PAM.
For a given data rate, M-ary PAM (M>2) requires less bandwidth than
binary PCM.
For a given average pulse power, binary PCM is easier to detect than M-
ary PAM (M>2).

26 : “shiv rpi”
PAM example: Binary vs 8-ary

27 : “shiv rpi”
Example of M-ary PAM
Assuming real time Tx and equal energy per Tx data bit for
binary-PAM and 4-ary PAM:
• 4-ary: T=2Tb and Binary: T=Tb
• Energy per symbol in binary-PAM: A = 10B

2 2
Binary PAM 4-ary PAM

(rectangular pulse) (rectangular pulse)
3B
A.
‘11’
‘1’ B
T
T T ‘01’
T ‘00’ T T
-B
‘0’ ‘10’
-A. -3B
28 : “shiv rpi”
Other PCM waveforms: Examples
PCM waveforms category:

Nonreturn-to-zero (NRZ) Phase encoded
Return-to-zero (RZ) Multilevel binary
+V
1 0 1 1 0 +V
1 0 1 1 0
NRZ-L -V Manchester -V
Unipolar-RZ +V Miller +V
0 -V
+V +V
Bipolar-RZ 0 Dicode NRZ 0
-V -V
0 T 2T 3T 4T 5T 0 T 2T 3T 4T 5T

29 : “shiv rpi”
PCM waveforms: Selection Criteria
Criteria for comparing and selecting PCM waveforms:
Spectral characteristics (power spectral density and
bandwidth efficiency)
Bit synchronization capability
Error detection capability
Interference and noise immunity
Implementation cost and complexity

30 : “shiv rpi”
Summary: Baseband Formatting and transmission
Digital info. Bit stream Pulse waveforms
(Data bits) (baseband signals)
Textual Format
source info.
Pulse
Analog Sample Quantize Encode modulate
info.
Sampling at rate Encoding each q. value to

f s = 1 / Ts l = log 2 L bits
(sampling time=Ts) (Data bit duration Tb=Ts/l)
Quantizing each sampled Mapping every m = log 2 M data bits to a

value to one of the symbol out of M symbols and transmitting
L levels in quantizer. a baseband waveform with duration T
Information (data- or bit-) rate: Rb = 1 / Tb [bits/sec]

Symbol rate : R = 1 / T [symbols/sec]
Rensselaer Polytechnic Institute Rb = mR Shivkumar Kalyanaraman

31 : “shiv rpi”
Receiver Structure & Matched Filter

32 : “shiv rpi”
Demodulation and detection
mi Pulse g i (t ) Bandpass si (t ) M-ary modulation
Format
modulate modulate i = 1, K, M
channel
transmitted symbol hc (t )
estimated symbol n(t )
Demod.
Format Detect
m̂i z (T ) & sample r (t )
Major sources of errors:

Thermal noise (AWGN)
disturbs the signal in an additive fashion (Additive)
has flat spectral density for all frequencies of interest (White)
is modeled by Gaussian random process (Gaussian Noise)
Inter-Symbol Interference (ISI)
Due to the filtering effect of transmitter, channel and receiver, symbols
are “smeared”. Shivkumar Kalyanaraman
33 : “shiv rpi”
Impact of AWGN

34 : “shiv rpi”
Impact of AWGN & Channel Distortion
hc (t ) = δ (t ) − 0.5δ (t − 0.75T )

35 : “shiv rpi”
Receiver job
Demodulation and sampling:
Waveform recovery and preparing the received signal for
detection:
Improving the signal power to the noise power (SNR)
using matched filter (project to signal space)
Reducing ISI using equalizer (remove channel distortion)
Sampling the recovered waveform
Detection:
Estimate the transmitted symbol based on the received sample

36 : “shiv rpi”
Receiver structure
Step 1 – waveform to sample transformation Step 2 – decision making
Demodulate & Sample Detect
z (T ) Threshold m̂i
r (t ) Frequency Receiving Equalizing
comparison
down-conversion filter filter
For bandpass signals Compensation for

channel induced ISI
Received waveform Baseband pulse

Baseband pulse Sample
(possibly distorted)
(test statistic)

37 : “shiv rpi”
Baseband vs Bandpass
Bandpass model of detection process is equivalent to baseband
model because:
The received bandpass waveform is first transformed to a
baseband waveform.
Equivalence theorem:
Performing bandpass linear signal processing followed by
heterodying the signal to the baseband, …
… yields the same results as …
… heterodying the bandpass signal to the baseband ,
followed by a baseband linear signal processing.

38 : “shiv rpi”
Steps in designing the receiver
Find optimum solution for receiver design with the

following goals:
1. Maximize SNR: matched filter
2. Minimize ISI: equalizer
Steps in design:
Model the received signal
Find separate solutions for each of the goals.
First, we focus on designing a receiver which maximizes the

SNR: matched filter

39 : “shiv rpi”
Receiver filter to maximize the SNR
Model the received signal
si (t ) hc (t ) r (t ) r (t ) = si (t ) ∗h c (t ) + n(t )
n(t )
AWGN
Simplify the model (ideal channel assumption):

Received signal in AWGN
Ideal channels
hc (t ) = δ (t )
si (t ) r (t ) r (t ) = si (t ) + n(t )
n(t )
AWGN

40 : “shiv rpi”
Problem:
Matched Filter Receiver
Design the receiver filter h(t ) such that the SNR is
maximized at the sampling time when si (t ), i = 1,..., M
is transmitted.
Solution:
The optimum filter, is the Matched filter, given by
h ( t ) = hopt ( t ) = s i (T − t )
*
H ( f ) = H opt ( f ) = S i ( f ) exp( − j 2π fT )
*
which is the time-reversed and delayed version of the conjugate of the

transmitted signal
si (t ) h(t ) = hopt (t )
0 T t 0 T t
41 : “shiv rpi”
Correlator Receiver
The matched filter output at the sampling time, can be realized
as the correlator output.
Matched filtering, i.e. convolution with si*(T-τ) simplifies to
integration w/ si*(τ), i.e. correlation or inner product!
z (T ) = hopt (T ) ∗ r (T )
T
= ∫ r (τ )s i (τ ) d τ =< r ( t ), s ( t ) >
*
Recall: correlation operation is the projection of the received

signal onto the signal space!
Key idea: Reject the noise (N) outside this space as irrelevant:
=> maximize S/N Shivkumar Kalyanaraman
42 : “shiv rpi”
Irrelevance Theorem: Noise Outside Signal Space
Noise PSD is flat (“white”) => total noise power

infinite across the spectrum.
We care only about the noise projected in the finite
signal dimensions (eg: the bandwidth of interest).
43 : “shiv rpi”
Aside: Correlation Effect
Correlation is a maximum when two signals are similar in shape, and are in phase (or
'unshifted' with respect to each other).
Correlation is a measure of the similarity between two signals as a function of time shift
(“lag”, τ ) between them
When the two signals are similar in shape and unshifted with respect to each other, their
product is all positive. This is like constructive interference,
The breadth of the correlation function - where it has significant value - shows for how long
the signals remain similar.
44 : “shiv rpi”
Aside: Autocorrelation

45 : “shiv rpi”
Aside: Cross-Correlation &Radar
Figure: shows how the signal can be located within the noise.
A copy of the known reference signal is correlated with the unknown signal.
The correlation will be high when the reference is similar to the unknown signal.
A large value for correlation shows the degree of confidence that the reference signal is
detected.
The large value of the correlation indicates when the reference signal occurs.
46 : “shiv rpi”
• A copy of a known reference signal is correlated with the unknown signal.
• The correlation will be high if the reference is similar to the unknown signal.
• The unknown signal is correlated with a number of known reference functions.
• A large value for correlation shows the degree of similarity to the reference.
• The largest value for correlation is the most likely match.
• Same principle in communications: reference signals corresponding to

symbols
• The ideal communications channel may have attenuated, phase shifted the
reference signal, and added noise

47 : “shiv
Source: Bores Signal rpi”
Processing
Matched Filter: back to cartoon…
Consider the received signal as a vector r, and the transmitted signal vector as s
Matched filter “projects” the r onto signal space spanned by s (“matches” it)
Filtered signal can now be safely sampled by the receiver at the correct sampling instants,
resulting in a correct interpretation of the binary message
Matched filter is the filter that maximizes the signal-to-noise ratio it can be
shown that it also minimizes the BER: it is a simple projection operation
48 : “shiv rpi”
Example of matched filter (real signals)
y (t ) = si (t ) ∗h opt (t )
si (t ) h opt (t ) A2
A A
T T
T t T t 0 T 2T t
y (t ) = si (t ) ∗h opt (t )
si (t ) h opt (t ) A2
A A
T T
T/2 T t T/2 T t 0 T/2 T 3T/2 2T t

−A −A − A2
2
T T

49 : “shiv rpi”
Properties of the Matched Filter
1. The Fourier transform of a matched filter output with the matched signal as input
is, except for a time delay factor, proportional to the ESD of the input signal.
Z ( f ) =| S ( f ) |2 exp(− j 2πfT )
2. The output signal of a matched filter is proportional to a shifted version of the
autocorrelation function of the input signal to which the filter is matched.
z (t ) = Rs (t − T ) ⇒ z (T ) = Rs (0) = Es
3. The output SNR of a matched filter depends only on the ratio of the signal energy
to the PSD of the white noise at the filter input.
⎛S⎞ Es
max⎜ ⎟ =
⎝ N ⎠T N 0 / 2
4. Two matching conditions in the matched-filtering operation:
spectral phase matching that gives the desired output peak at time T.
spectral amplitude matching that gives optimum SNR to the peak value.

50 : “shiv rpi”
Implementation of matched filter receiver
Bank of M matched filters
z1 (T )
s (T − t ) ⎡ z1 ⎤
*
1
Matched filter output:
r (t ) ⎢ M ⎥=z z Observation
⎢ ⎥ vector
⎢⎣zM ⎥⎦
sM (T − t )
*
zM (T )
zi = r (t ) ∗ s ∗i (T − t ) i = 1,..., M
z = ( z1 (T ), z 2 (T ),..., z M (T )) = ( z1 , z 2 ,..., z M )
Note: we are projecting along the basis directions of the signal space
51 : “shiv rpi”
Implementation of correlator receiver
Bank of M correlators
s ∗1 (t )
T z1 (T )
∫0
⎡ z1 ⎤ Correlators output:
r (t ) ⎢M⎥ =z
z Observation
⎢ ⎥
∗
s M (t ) vector
T ⎢⎣zM ⎥⎦
∫0 z M (T )
z = ( z1 (T ), z 2 (T ),..., z M (T )) = ( z1 , z 2 ,..., z M )
T
zi = ∫ r (t )si (t )dt i = 1,..., M
0
Note: In previous slide we “filter” i.e. convolute inShivkumar
the boxesKalyanaraman
shown.
52 : “shiv rpi”
Implementation example of matched filter receivers
s1 (t )
A
Bank of 2 matched filters
T
0 T t A z1 (T )
T ⎡z1 ⎤
r (t )
0 T
⎢ ⎥ =z
z
s2 (t ) ⎢ ⎥
⎢⎣z2 ⎥⎦
0 T
z 2 (T )
0 T t
−A −A
T T

53 : “shiv rpi”
Matched Filter: Frequency domain View
Simple Bandpass Filter:

excludes noise, but misses some signal power

54 : “shiv rpi”
Matched Filter: Frequency Domain View (Contd)
Multi-Bandpass Filter: includes more signal power, but adds more noise also!
Matched Filter: includes more signal power, weighted according to size

=> maximal noise rejection!

55 : “shiv rpi”
Maximal Ratio Combining (MRC) viewpoint
Generalization of this f-domain picture, for combining
multi-tap signal
Weight each branch
SNR:
56 : “shiv rpi”
Examples of matched filter output for bandpass
modulation schemes

57 : “shiv rpi”
Signal Space Concepts

58 : “shiv rpi”
Signal space: Overview
What is a signal space?
Vector representations of signals in an N-dimensional orthogonal space
Why do we need a signal space?

It is a means to convert signals to vectors and vice versa.
It is a means to calculate signals energy and Euclidean distances
between signals.
Why are we interested in Euclidean distances between signals?

For detection purposes: The received signal is transformed to a received
vectors.
The signal which has the minimum distance to the received signal is
estimated as the transmitted signal.
59 : “shiv rpi”
Schematic example of a signal space
ψ 2 (t )
s1 = (a11 , a12 )
ψ 1 (t )
z = ( z1 , z 2 )
s 3 = (a31 , a32 )
s 2 = (a21 , a22 )
s1 (t ) = a11ψ 1 (t ) + a12ψ 2 (t ) ⇔ s1 = (a11 , a12 )
Transmitted signal
s2 (t ) = a21ψ 1 (t ) + a22ψ 2 (t ) ⇔ s 2 = (a21 , a22 )
alternatives
s3 (t ) = a31ψ 1 (t ) + a32ψ 2 (t ) ⇔ s 3 = (a31 , a32 )
Received signal at
z (t ) = z1ψ 1 (t ) + z2ψ 2 (t ) ⇔ z = ( z1 , z 2 )
matched
Rensselaerfilter output
Polytechnic Institute Shivkumar Kalyanaraman
60 : “shiv rpi”
Signal space
To form a signal space, first we need to know the
inner product between two signals (functions):
Inner (scalar) product:
∞
< x(t ), y (t ) >= ∫ x t y (t )dt
*
( )
−∞
= cross-correlation between x(t) and y(t)
Properties of inner product:

< ax (t ), y (t ) >= a < x(t ), y (t ) >
< x(t ), ay (t ) >= a * < x(t ), y (t ) >
< x(t ) + y (t ), z (t ) >=< x(t ), z (t ) > + < y (t ), z (t ) >
61 : “shiv rpi”
Signal space …
The distance in signal space is measure by calculating the norm.
What is norm?
Norm of a signal:
∞
x(t ) = < x(t ), x(t ) > = ∫ x(t ) dt = E x
2
−∞
= “length” of x(t)
ax(t ) = a x(t )
Norm between two signals:
d x , y = x(t ) − y (t )
We refer to the norm between two signals as the Euclidean distance between
two signals.
62 : “shiv rpi”
Example of distances in signal space
ψ 2 (t )
s1 = (a11 , a12 )
E1 d s1 , z
ψ 1 (t )
E3 z = ( z1 , z 2 )
d s3 , z E2 d s2 , z
s 3 = (a31 , a32 )
s 2 = (a21 , a22 )
The Euclidean distance between signals z(t) and s(t):

d si , z = si (t ) − z (t ) = (ai1 − z1 ) 2 + (ai 2 − z 2 ) 2
i = 1,2,3
63 : “shiv rpi”
Orthogonal signal space
N-dimensional orthogonal signal space is characterized by N
linearly independent functions {ψ j (t )}j =1 called basis functions.
N
The basis functions must satisfy the orthogonality condition
T
0≤t ≤T
< ψ i (t ),ψ j (t ) >= ∫ψ i (t )ψ (t )dt = K iδ ji
*
j
0
j , i = 1,..., N
where
⎧1 → i = j
δ ij = ⎨
⎩0 → i ≠ j
If all K i = 1 , the signal space is orthonormal.

Constructing Orthonormal basis from non-orthonormal set of vectors:
Gram-Schmidt procedure

64 : “shiv rpi”
Example of an orthonormal bases
Example: 2-dimensional orthonormal signal space
⎧ 2
ψ
⎪ 1 (t ) = cos(2πt / T ) 0≤t <T ψ 2 (t )
⎪ T
⎨
⎪ψ (t ) = 2 sin( 2πt / T ) 0≤t <T
⎪⎩ 2 T
T
ψ 1 (t )
0
< ψ 1 (t ),ψ 2 (t ) >= ∫ψ 1 (t )ψ 2 (t )dt = 0
0
ψ 1 (t ) = ψ 2 (t ) = 1
Example: 1-dimensional orthonornal signal space

ψ 1 (t )
1 ψ 1 (t ) = 1
T ψ 1 (t )
0
0 T t
65 : “shiv rpi”
Sine/Cosine Bases: Note!
Approximately orthonormal!
These are the in-phase & quadrature-phase dimensions of

complex baseband equivalent representations.
66 : “shiv rpi”
Example: BPSK
Note: two symbols, but only one dimension in BPSK.

67 : “shiv rpi”
Signal space …
Any arbitrary finite set of waveforms {si (t )}iM=1
where each member of the set is of duration T, can be expressed
as a linear combination of N orthogonal waveforms
where {ψ .(t )}N N≤M
j j =1
N
si (t ) = ∑ aijψ j (t ) i = 1,..., M
j =1 N≤M
where
T
1 1 j = 1,..., N
aij =
Kj
< si (t ),ψ j (t ) >= ∫
Kj 0
si (t )ψ *j (t )dt
i = 1,..., M
0≤t ≤T
N
Ei = ∑ K j aij
2
s i = (ai1 , ai 2 ,..., aiN )
j =1
Vector representation of waveform Waveform energy
68 : “shiv rpi”
Signal space …
N
s i (t ) = ∑ a ijψ j (t ) s i = (ai1 , ai 2 ,..., aiN )
j =1
Waveform to vector conversion Vector to waveform conversion
ψ 1 (t ) ψ 1 (t )
T ai1 ai1
∫ ⎡ai1 ⎤ ⎡ai1 ⎤
⎢M⎥
0 sm
si (t ) ⎢ M ⎥ = sm si (t )
ψ N (t ) ⎢ ⎥ ⎢ ⎥ ψ N (t )
T ⎢⎣aiN ⎥⎦ ⎢⎣aiN ⎥⎦
∫0 aiN aiN

69 : “shiv rpi”
Example of projecting signals to an
orthonormal signal space
ψ 2 (t )
s1 = (a11 , a12 )
ψ 1 (t )
s 3 = (a31 , a32 )
s 2 = (a21 , a22 )
s1 (t ) = a11ψ 1 (t ) + a12ψ 2 (t ) ⇔ s1 = (a11 , a12 )

Transmitted signal
alternatives s2 (t ) = a21ψ 1 (t ) + a22ψ 2 (t ) ⇔ s 2 = (a21 , a22 )
s3 (t ) = a31ψ 1 (t ) + a32ψ 2 (t ) ⇔ s 3 = (a31 , a32 )
T
Rensselaer Polytechnic Institute aij = ∫ si (t )ψ j (t )dt j = 1,..., N 0 ≤ t ≤ TKalyanaraman

i = 1,..., M Shivkumar
0
70 : “shiv rpi”
Matched filter receiver (revisited)
(note: we match to the basis directions)
Bank of N matched filters
z1
ψ (T − t )
∗
⎡ z1 ⎤
1 Observation
vector
r (t )
⎢ ⎥=z z
⎢ ⎥
ψ ∗ N (T − t ) ⎢⎣zN ⎥⎦
zN
N
si (t ) = ∑ aijψ j (t ) i = 1,..., M
j =1
z = ( z1 , z 2 ,..., z N ) N≤M
z j = r (t ) ∗ψ j (T − t ) j = 1,..., N
71 : “shiv rpi”
Correlator receiver (revisited)
Bank of N correlators
ψ 1 (t )
T z1
∫0
⎡ r1 ⎤
r (t ) ⎢M⎥ =z
z Observation
ψ N (t ) ⎢ ⎥ vector
T ⎢⎣rN ⎥⎦
∫0 zN
N
si (t ) = ∑ aijψ j (t ) i = 1,..., M
j =1
z = ( z1 , z 2 ,..., z N ) N≤M
T
z j = ∫ r (t )ψ j (t )dt j = 1,..., N
0 Institute
Rensselaer Polytechnic Shivkumar Kalyanaraman
72 : “shiv rpi”
Example of matched filter receivers using
basic functions
s1 (t ) s2 (t ) ψ 1 (t )
A 1
T T
0 T t
0 T t −A 0 T t
T
1 matched filter
ψ 1 (t )
r (t )
[z1] = z
1
T z1 z
0 T t
Number of matched filters (or correlators) is reduced by 1 compared to

using matched filters (correlators) to the transmitted signal!
73 : “shiv rpi”
White noise in Orthonormal Signal Space
AWGN n(t) can be expressed as
n(t ) = nˆ (t ) + n~ (t )
Noise outside on the signal space
Noise projected on the signal space
(irrelevant)
(colored):impacts the detection process.
N
nˆ (t ) = ∑ n jψ j (t ) nˆ (t )
Vector representation of
j =1
n j =< n(t ),ψ j (t ) > j = 1,..., N n = (n1 , n2 ,..., nN )
< n~ (t ),ψ (t ) >= 0

j
j = 1,..., N {n } N
independent zero-mean
j j =1
Gaussain random variables with
variance var(n j ) = N 0 / 2

74 : “shiv rpi”
Detection: Maximum Likelihood &
Performance Bounds

75 : “shiv rpi”
Detection of signal in AWGN
Detection problem:
Given the observation vector z , perform a mapping from z
to an estimate m̂ of the transmitted symbol, mi , such that
the average probability of error in the decision is
minimized.
n
mi si z m̂
Modulator Decision rule

76 : “shiv rpi”
Statistics of the observation Vector
AWGN channel model: z = si + n
Signal vector s i = (ai1 , ai 2 ,..., aiN ) is deterministic.
Elements of noise vector n = ( n1 , n2 ,..., nN ) are i.i.d Gaussian
random variables with zero-mean and variance N 0 / 2 . The
noise vector pdf is
1 ⎛ n
2
⎞
pn (n) = exp⎜− ⎟
(πN 0 )N / 2 ⎜⎝ N 0 ⎟⎠
The elements of observed vector z = ( z1 , z 2 ,..., z N ) are
independent Gaussian random variables. Its pdf is
1 ⎛ z − s
2
⎞
pz ( z | s i ) = exp⎜ − i ⎟
(πN 0 )N /2
⎜
⎝ N 0
⎟
⎠
77 : “shiv rpi”
Detection
Optimum decision rule (maximum a posteriori probability):
Set mˆ = mi if
Pr( mi sent | z ) ≥ Pr(mk sent | z ), for all k ≠ i
where k = 1,..., M .
Applying Bayes’ rule gives:

Set mˆ = mi if
pz (z | mk )
pk , is maximum for all k = i
pz ( z )
78 : “shiv rpi”
Detection …
Partition the signal space into M decision regions,
such that Z1 ,..., Z M
Vector z lies inside region Z i if

pz (z | mk )
ln[ pk ], is maximum for all k = i.
pz ( z )
That means
mˆ = mi

79 : “shiv rpi”
Detection (ML rule)
For equal probable symbols, the optimum decision rule
(maximum posteriori probability) is simplified to:
Set mˆ = mi if
pz (z | mk ), is maximum for all k = i
or equivalently:
Set mˆ = mi if
ln[ pz (z | mk )], is maximum for all k = i
which is known as maximum likelihood.

80 : “shiv rpi”
Detection (ML)…
Partition the signal space into M decision regions, Z1 ,..., Z M
Restate the maximum likelihood decision rule as follows:

ln[ pz (z | mk )], is maximum for all k = i
That means
mˆ = mi

z − s k , is minimum for all k = i

81 : “shiv rpi”
Schematic example of ML decision regions
ψ 2 (t )
Z2
s2
Z1
s3 s1
Z3 ψ 1 (t )
s4
Z4
82 : “shiv rpi”
Probability of symbol error
Erroneous decision: For the transmitted symbol s i or equivalently signal
vector mi , an error in decision occurs if the observation vector z does not
fall inside region Z i.
Probability of erroneous decision for a transmitted symbol
Pr(mˆ ≠ mi ) = Pr(mi sent)Pr(z does not lie inside Z i mi sent)
Probability of correct decision for a transmitted symbol
Pr(mˆ = mi ) = Pr(mi sent)Pr(z lies inside Z i mi sent)
Pc (mi ) = Pr(z lies inside Z i mi sent) = ∫ p (z | m )dz

Zi
z i
Pe (mi ) = 1 − Pc (mi )
83 : “shiv rpi”
Example for binary PAM
pz (z | m2 ) pz (z | m1 )
s2 s1
ψ 1 (t )
− Eb 0 Eb
⎛ s1 − s 2 / 2 ⎞
Pe (m1 ) = Pe (m2 ) = Q⎜ ⎟
⎜ N /2 ⎟
⎝ 0 ⎠
⎛ 2 Eb ⎞
PB = PE (2) = Q⎜⎜ ⎟
⎟
⎝ N0 ⎠
84 : “shiv rpi”
Average prob. of symbol error …
Average probability of symbol error :
M
PE ( M ) = ∑ Pr ( mˆ ≠ m )
i =1
i
For equally probable symbols:
M M
1 1
PE ( M ) =
M
∑
i =1
Pe (mi ) = 1 −
M
∑ P (m )
i =1
c i
M
1
= 1−
M
∑ ∫ p (z | m )dz
i =1 Z i
z i

85 : “shiv rpi”
Union bound
Union bound
The probability of a finite union of events is upper bounded
by the sum of the probabilities of the individual events.
M M M
1
Pe (mi ) ≤ ∑ P2 (s k , s i ) PE ( M ) ≤ ∑∑ P (s 2 k , si )
k =1 M i =1 k =1
k ≠i k ≠i

86 : “shiv rpi”
Example of union bound
ψ2
Z2 r Z1
Pe (m1 ) = ∫ p (r | m )dr
r
Z 2 ∪Z3 ∪Z 4
1 s2 s1
ψ1
Union bound: s3 s4
4 Z3 Z4
Pe (m1 ) ≤ ∑ P2 (s k , s1 )
k =2
ψ2 ψ2 ψ2
A2 r r r
s2 s1 s2 s1 s2 s1
ψ1 ψ1 ψ1
s3 s4 s3 s4 s3 s4
A3 A4
P2 (s 2 , s1 ) = ∫ p (r | m )dr
A
r 1 P2 (s 3 , s1 ) = ∫ p (r | m )dr
A3
r 1 P2 (s 4 , s1 ) = ∫ pr (r | m1 )dr
2 A4
Shivkumar Kalyanaraman
87 : “shiv rpi”
Upper bound based on minimum distance
P2 (s k , s i ) = Pr(z is closer to s k than s i , when s i is sent)
∞
1 u2 ⎛ d ik / 2 ⎞
= ∫ exp(− )du =Q⎜ ⎟
πN 0 N0 ⎜ N /2 ⎟
d ik ⎝ 0 ⎠
d ik = s i − s k
1 M M ⎛ d min / 2 ⎞
PE ( M ) ≤
M
∑∑ P2 (s k , s i ) ≤ ( M − 1)Q⎜
⎜ N /2 ⎟
⎟
i =1 k =1
k ≠i
⎝ 0 ⎠
d min = min d ik
Minimum distance in the signal space: i ,k
i≠k

88 : “shiv rpi”
Example of upper bound on av. Symbol error prob.
based on union bound
s i = Ei = Es , i = 1,...,4 ψ 2 (t )
d i ,k = 2 Es
i≠k Es s2
d min = 2 Es d 2,3 d1, 2
s3 s1
ψ 1 (t )
− Es Es
d 3, 4 d1, 4
s4
− Es

89 : “shiv rpi”
90 : “shiv rpi”
Eb/No figure of merit in digital
communications
SNR or S/N is the average signal power to the average noise
power. SNR should be modified in terms of bit-energy in
digital communication, because:
Signals are transmitted within a symbol duration and hence,
are energy signal (zero power).
A metric at the bit-level facilitates comparison of different
DCS transmitting different number of bits per symbol.
Eb STb S W Rb : Bit rate

= =
N 0 N / W N Rb W : Bandwidth
Note: S/N = Eb/No x spectral efficiency

91 : “shiv rpi”
Example of Symbol error prob. For PAM signals
Binary PAM
s2 s1
ψ 1 (t )
− Eb 0 Eb
4-ary PAM
s4 s3 s2 s1
Eb E 0 E Eb ψ 1 (t )
−6 −2 b 2 b 6
5 5 5 5
ψ1 (t )
1
T
0 T t
92 : “shiv rpi”
Maximum Likelihood (ML) Detection: Vector Case
Nearest Neighbor Rule:
By the isotropic property of the Gaussian noise, we expect the error
probability to be the same for both the transmit symbols uA, uB.
Error probability:
Project the received vector y along the

difference vector direction uA- uB is a
“sufficient statistic”.
Noise outside these finite dimensions is

irrelevant for detection. (rotational
invariance of detection problem)
ps: Vector norm is a natural extension of “magnitude” or length
93 : “shiv rpi”
Extension to M-PAM (Multi-Level Modulation)
Note: h refers to the constellation shape/direction
MPAM:

94 : “shiv rpi”
Complex Vector Space Detection
Error probability:
Note: Instead of vT, use v* for complex vectors (“transpose and conjugate”)
for inner products…
95 : “shiv rpi”
Complex
Detection:
Summary

96 : “shiv rpi”
Detection Error => BER
If the bit error is i.i.d (discrete memoryless channel) over the sequence of bits,
then you can model it as a binary symmetric channel (BSC)
BER is modeled as a uniform probability f
As BER (f) increases, the effects become increasingly intolerable
f tends to increase rapidly with lower SNR: “waterfall” curve (Q-function)

97 : “shiv rpi”
SNR vs BER: AWGN vs Rayleigh
Need
diversity
techniques
to deal with
Rayleigh
(even 1-tap,
flat-fading)!
Observe the “waterfall” like characteristic (essentially plotting the Q(x) function)!
Telephone lines: SNR = 37dB, but low b/w (3.7kHz)
Wireless: Low SNR = 5-10dB, higher bandwidth (upto 10 Mhz, MAN, and 20Mhz LAN)
Optical fiber comm: High SNR, high bandwidth ! But cant process w/ complicated codes,
signal processing etc
98 : “shiv rpi”
Better performance through diversity
Diversity Ù the receiver is provided with multiple copies

of the transmitted signal. The multiple signal copies
should experience uncorrelated fading in the channel.
In this case the probability that all signal copies fade
simultaneously is reduced dramatically with respect to
the probability that a single copy experiences a fade.
As a rough rule:
1 Diversity
Diversity of
Pe is proportional to of
γ 0L L:th
L:th order
order
BER
BER Average
Average SNR
SNR

99 : “shiv rpi”
BER vs. SNR (diversity effect)
BER
( = Pe )
Flat fading channel,

Rayleigh fading,
AWGN L=1
channel
(no fading)
SNR (= γ0)
L=4 L=3 L=2
We will explore this story later… slide set part II

100 : “shiv rpi”
Modulation Techniques

What is Modulation?
Encoding information in a manner suitable for
transmission.
Translate baseband source signal to bandpass signal
Bandpass signal: “modulated signal”
How?
Vary amplitude, phase or frequency of a carrier
Demodulation: extract baseband message from carrier

Digital vs Analog Modulation
Cheaper, faster, more power efficient
Higher data rates, power error correction, impairment
resistance:
Using coding, modulation, diversity
Equalization, multicarrier techniques for ISI
mitigation
More efficient multiple access strategies, better
security: CDMA, encryption etc

Goals of Modulation Techniques
• High Bit Rate
• High Spectral Efficiency (max Bps/Hz)
• High Power Efficiency (min power to achieve a target BER)
• Low-Cost/Low-Power Implementation
• Robustness to Impairments

Modulation: representation
• Any modulated signal can be represented as
s(t) = A(t) cos [ωct + φ(t)]
amplitude phase or frequency
s(t) = A(t) cos φ(t) cos ωct - A(t) sin φ(t) sin ωct
in-phase quadrature
• Linear versus nonlinear modulation ⇒ impact on spectral efficiency

Linear: Amplitude or phase
Non-linear: frequency: spectral broadening
• Constant envelope versus non-constant envelope
⇒ hardware implications with impact on power efficiency
(=> reliability: i.e. target BER at lower SNRs)
Complex Vector Spaces: Constellations
MPSK
Circular Square
Each signal is encoded (modulated) as a vector in a signal space

Linear Modulation Techniques
s(t) = [ Σ an g (t-nT)]cos ωct - [ Σ bn g (t-nT)] sin ωct
n n
I(t), in-phase Q(t), quadrature
LINEAR MODULATIONS
Square M-ARY QUADRATURE M-ARY PHASE Circular

Constellations AMPLITUDE MOD. SHIFT KEYING Constellations
(M-QAM) (M-PSK)
M≠4 M=4 M≠4

(4-QAM = 4-PSK)
CONVENTIONAL OFFSET DIFFERENTIAL

4-PSK 4-PSK 4-PSK
(QPSK) (OQPSK) (DQPSK, π/4-DQPSK)

M-PSK and M-QAM
M-PSK (Circular Constellations) M-QAM (Square Constellations)

bn bn
4-PSK
16-QAM
16-PSK
4-PSK
an an
Tradeoffs
– Higher-order modulations (M large) are more spectrally
efficient but less power efficient (i.e. BER higher).
– M-QAM is more spectrally efficient than M-PSK but
also more sensitive to system nonlinearities.

Bandwidth vs. Power Efficiency
MPSK:
MQAM:
MFSK:

Source: Rappaport book, chap 6 : “shiv rpi”
109
MPAM & Symbol Mapping
Binary PAM
Note: the average energy s2 s1
ψ 1 (t )
per-bit is constant − Eb 0 Eb
Gray coding used for

4-ary PAM
mapping bits to symbols s4 s3 s2 s1
Why? Most likely error is to Eb Eb 0 Eb Eb ψ 1 (t )

−6 −2 2 6
confuse with neighboring 5 5 5 5
symbol.
Make sure that the
Gray coding
00 01 11 10
neighboring symbol has only 4-ary PAM
s4 s3 s2 s1
1-bit difference (hamming
distance = 1) Eb E 0 E Eb ψ 1 (t )
−6 −2 b 2 b 6
5 5 5 5

MPAM: Details
Unequal energies/symbol:

Decision111
Regions : “shiv rpi”
M-PSK (Circular Constellations)
MPSK: bn
4-PSK
16-PSK
an
Constellation points:
Equal energy in all signals:

01
01 00
Gray coding 00
11
10
11 Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute 10
MPSK: Decision Regions & Demod’ln
Z3
Z2 Z4
Z2
Z3 Z1 Z5 Z1
Z6 Z8
Z4 Z7
4PSK 8PSK
4PSK: 1 bit/complex dimension or 2 bits/symbol
m=1
si(t) + n(t) Z1: r > 0 m = 0 or 1

X g(Tb - t)
Z2: r ≤ 0
m=0
cos(2πfct)
Coherent Demodulator for BPSK.
M-QAM (Square Constellations)
MQAM: bn
16-QAM
4-PSK
an
Unequal symbol energies:

Z1 Z2 Z3 Z4
MQAM with square constellations of size is L2
equivalent to MPAM modulation with
constellations of size L on each of the in-phase Z5 Z6 Z7 Z8
and quadrature signal components
For square constellations it takes approximately Z9 Z10 Z11 Z12

6 dB more power to send an additional 1
bit/dimension or 2 bits/symbol while
maintaining the same minimum distance Z13 Z14 Z15 Z16
between constellation points
Hard to find a Gray code mapping where all

adjacent symbols differ by a single bit 16QAM: Decision Regions
Non-Coherent Modulation: DPSK
Information in MPSK, MQAM carried in signal phase.
Requires coherent demodulation: i.e. phase of the transmitted signal carrier φ0 must
be matched to the phase of the receiver carrier φ
More cost, susceptible to carrier phase drift.
Harder to obtain in fading channels
Differential modulation: do not require phase reference.

More general: modulation w/ memory: depends upon prior symbols transmitted.
Use prev symbol as the a phase reference for current symbol
Info bits encoded as the differential phase between current & previous symbol
Less sensitive to carrier phase drift (f-domain) ; more sensitive to doppler
effects: decorrelation of signal phase in time-domain

Differential Modulation (Contd)
DPSK: Differential BPSK
A 0 bit is encoded by no change in phase, whereas a 1 bit is encoded as a
phase change of π.
If symbol over time [(k−1)Ts, kTs) has phase θ(k − 1) = ejθi , θi = 0, π,
then to encode a 0 bit over [kTs, (k + 1)Ts), the symbol would have
phase: θ(k) = ejθi and…
… to encode a 1 bit the symbol would have
phase θ(k) = ej(θi+π).
DQPSK: gray coding:

Quadrature Offset
Phase transitions of 180o can cause large amplitude transitions (through
zero point).
Abrupt phase transitions and large amplitude variations can be distorted
by nonlinear amplifiers and filters
Avoided by offsetting the quadrature branch pulse g(t) by half a symbol

period
Usually abbreviated as O-MPSK, where the O indicates the offset
QPSK modulation with quadrature offset is referred to as O-QPSK
O-QPSK has the same spectral properties as QPSK for linear
amplification,..
… but has higher spectral efficiency under nonlinear amplification,
since the maximum phase transition of the signal is 90 degrees
Another technique to mitigate the amplitude fluctuations of a 180 degree

phase shift used in the IS-54 standard for digital cellular is π/4-QPSK
Maximum phase transition of 135 degrees, versus 90 degrees for offset
QPSK and 180 degrees for QPSK
Offset QPSK waveforms

Frequency Shift Keying (FSK)
• Continuous Phase FSK (CPFSK)
– digital data encoded in the frequency shift
– typically implemented with frequency
modulator to maintain continuous phase
s(t) = A cos [ωct + 2 πkf ∫ d(τ) dτ]
t
−∞
– nonlinear modulation but constant-envelope

• Minimum Shift Keying (MSK)
– minimum bandwidth, sidelobes large
– can be implemented using I-Q receiver
• Gaussian Minimum Shift Keying (GMSK)

– reduces sidelobes of MSK using a premodulation filter
– used by RAM Mobile Data, CDPD, and HIPERLAN

Minimum Shift Keying (MSK) spectra

Spectral Characteristics
10
0 QPSK/DQPSK
GMSK
Power Spectral Density (dB)
-20
-40
-60
B3-dBTb = 0.16
0.25
-80
1.0
-100 (MSK)
-120
0 0.5 1.0 1.5 2.0 2.5
Normalized Frequency (f-fc)Tb
10-1
Bit Error Probability (BER): AWGN
5
2
10-2
5 -3
For Pb = 10
2 DBPSK BPSK 6.5 dB
10-3 BPSK, QPSK QPSK 6.5 dB
5 DBPSK ~8 dB
Pb DQPSK ~9 dB
2
DQPSK
10-4
5 • QPSK is more spectrally efficient than BPSK
with the same performance.
2
10-5
• M-PSK, for M>4, is more spectrally efficient
but requires more SNR per bit.
5
2
• There is ~3 dB power penalty for differential
detection.
10-6
0 2 4 6 8 10 12 14
γb, SNR/bit, dB

Bit Error Probability (BER): Fading Channel
1
5
2
10-1
5
2
10-2
5 DBPSK
Pb • Pb is inversely proportion to the average SNR per bit.
2
BPSK
10-3 • Transmission in a fading environment requires about
5 AWGN 18 dB more power for Pb = 10-3.
2
10-4
5
2
10-5
0 5 10 15 20 25 30 35
γb, SNR/bit, dB
Bit Error Probability (BER): Doppler Effects
• Doppler causes an irreducible error floor when differential
detection is used ⇒ decorrelation of reference signal.
100
QPSK
DQPSK
10 -1
Rayleigh Fading
• The irreducible Pb depends on the data rate and the Doppler.
10 -2 For fD = 80 Hz,
data rate T Pbfloor
Pb 10
-3
10 kbps 10-4s 3x10-4
fDT=0.003
100 kbps 10-5s 3x10-6
No Fading 1 Mbps 10-6s 3x10-8
10 -4 0.002
The implication is that Doppler is not an issue for high-speed
0.001 wireless data.
10 -5 [M. D. Yacoub, Foundations of Mobile Radio Engineering ,
0 CRC Press, 1993]
10 -6
0 10 20 30 40 50 60
γb, SNR/bit, dB

Bit Error Probability (BER): Delay Spread
• ISI causes an irreducible error floor.
10-1
Coherent Detection
+ BPSK
QPSK
OQPSK Modulation
x MSK
• The rms delay spread imposes a limit on the
x
10-2 maximum bit rate in a multipath environment.
Irreducible Pb
x
For example, for QPSK,
x τ Maximum Bit Rate
+
Mobile (rural) 25 μsec 8 kbps
x
+ Mobile (city) 2.5 μsec 80 kbps
Microcells 500 nsec 400 kbps
+ Large Building 100 nsec 2 Mbps
10-3
+ [J. C.-I. Chuang, "The Effects of Time Delay Spread on Portable Radio
x Communications Channels with Digital Modulation,"
IEEE JSAC, June 1987]
10-4
10-2 10-1 100
rms delay spread τ
=
symbol period T
Summary of Modulation Issues
• Tradeoffs
– linear versus nonlinear modulation

– constant envelope versus non-constant
envelope
– coherent versus differential detection
– power efficiency versus spectral efficiency
• Limitations
– flat fading
– doppler
– delay spread
Pulse Shaping

Recall: Impact of AWGN only

Impact of AWGN & Channel Distortion
hc (t ) = δ (t ) − 0.5δ (t − 0.75T )
Multi-tap, ISI channel

ISI Effects: Band-limited Filtering of Channel
ISI due to filtering effect of the communications
channel (e.g. wireless channels)
Channels behave like band-limited filters
jθ c ( f )
Hc ( f ) = Hc ( f ) e
Non-constant amplitude Non-linear phase
Amplitude distortion Phase distortion

Inter-Symbol Interference (ISI)
ISI in the detection process due to the filtering effects
of the system
Overall equivalent system transfer function
H ( f ) = Ht ( f )H c ( f )H r ( f )
createsechoes and hence time dispersion
causes ISI at sampling time
ISI effect
z k = sk + nk + ∑ α i si
i≠k

Inter-symbol interference (ISI): Model
Baseband system model
x1 x2
zk
{xk } Tx filter Channel r (t ) Rx. filter {x̂k }
ht (t ) hc (t ) hr (t ) Detector
t = kT
T Ht ( f ) Hc ( f ) Hr ( f )
x3 T n(t )
Equivalent model
x1 x2
zk
{xk } Equivalent system z (t ) {x̂k }
h(t ) Detector
t = kT
T H( f )
x3 T nˆ (t )
filtered noise
H ( f ) = Ht ( f )H c ( f )H r ( f )
Nyquist bandwidth constraint
Nyquist bandwidth constraint (on equivalent system):
The theoretical minimum required system bandwidth to
detect Rs [symbols/s] without ISI is Rs/2 [Hz].
Equivalently, a system with bandwidth W=1/2T=Rs/2 [Hz]
can support a maximum transmission rate of 2W=1/T=Rs
[symbols/s] without ISI.
1 Rs Rs
= ≤W ⇒ ≥ 2 [symbol/s/Hz]
2T 2 W
Bandwidth efficiency, R/W [bits/s/Hz] :
An important measure in DCs representing data
throughput per hertz of bandwidth.
Showing how efficiently the bandwidth resources are used
by signaling techniques. Shivkumar Kalyanaraman

Equiv System: Ideal Nyquist pulse (filter)
Ideal Nyquist filter Ideal Nyquist pulse
H( f ) h(t ) = sinc(t / T )
T 1
0 f − 2T − T 0 T 2T t
−1 1
2T 2T
1
W=
Rensselaer Polytechnic Institute 2T Shivkumar Kalyanaraman
Nyquist pulses (filters)
Nyquist pulses (filters):
Pulses (filters) which result in no ISI at the sampling time.
Nyquist filter:
Its transfer function in frequency domain is obtained by
convolving a rectangular function with any real even-
symmetric frequency function
Nyquist pulse:
Its shape can be represented by a sinc(t/T) function multiply
by another time function.
Example of Nyquist filters: Raised-Cosine filter

Pulse shaping to reduce ISI
Goals and trade-off in pulse-shaping
Reduce ISI
Efficient bandwidth utilization
Robustness to timing error (small side lobes)

Raised Cosine Filter: Nyquist Pulse Approximation
| H ( f ) |=| H RC ( f ) | h(t ) = hRC (t )
1 r =0 1
r = 0. 5
0.5 0.5 r =1
r =1 r = 0.5
r =0
−1 − 3 −1 0 1 3 1 − 3T − 2T − T 0 T 2T 3T
T 4T 2T 2T 4T T
Rs
Baseband W sSB= (1 + r ) Passband W DSB= (1 + r ) Rs
2

Raised Cosine Filter
Raised-Cosine Filter
A Nyquist pulse (No ISI at the sampling time)
⎧1 for | f |< 2W0 − W

⎪⎪
2 ⎡ π | f | +W − 2W0 ⎤
H ( f ) = ⎨cos ⎢ ⎥ for 2W0 − W <| f |< W
⎪ ⎣4 W − W0 ⎦
⎪⎩0 for | f |> W
cos[2π (W − W0 )t ]
h(t ) = 2W0 (sinc(2W0t ))
1 − [4(W − W0 )t ]2
W − W0
Excess bandwidth: W − W Roll-off factor r =
W0
0 ≤ r ≤1
0

Pulse Shaping and Equalization Principles
No ISI at the sampling time
H RC ( f ) = H t ( f ) H c ( f ) H r ( f ) H e ( f )
Square-Root Raised Cosine (SRRC) filter and Equalizer
H RC ( f ) = H t ( f ) H r ( f )
Taking care of ISI
H r ( f ) = H t ( f ) = H RC ( f ) = H SRRC ( f ) caused by tr. filter
1
He ( f ) = Taking care of ISI
Hc ( f ) caused by channel

Pulse Shaping & Orthogonal Bases

Virtue of pulse shaping

PSD of a BPSK signal Shivkumar Kalyanaraman
Source: Rappaport book, chap 6 : “shiv rpi”
141
Example of pulse shaping
Square-root Raised-Cosine (SRRC) pulse shaping
Amp. [V]
Baseband tr. Waveform
Third pulse
t/T
First pulse
Second pulse
Data symbol

Example of pulse shaping …
Raised Cosine pulse at the output of matched filter
Amp. [V]
Baseband received waveform at

the matched filter output
(zero ISI)
t/T

Eye pattern
Eye pattern:Display on an oscilloscope which sweeps the system
response to a baseband signal at the rate 1/T (T symbol duration)
Distortion
due to ISI
Noise margin
amplitude scale
Sensitivity to
timing error
Timing jitter
timeShivkumar
scale Kalyanaraman
Example of eye pattern:
Binary-PAM, SRRC pulse
Perfect channel (no noise and no ISI)

Binary-PAM, SRRC pulse …
AWGN (Eb/N0=20 dB) and no ISI

Binary-PAM, SRRC pulse …
AWGN (Eb/N0=10 dB) and no ISI

Summary
Digital Basics
Modulation & Detection, Performance, Bounds
Modulation Schemes, Constellations
Pulse Shaping

Extra Slides

Bandpass Modulation: I, Q Representation

Analog: Frequency Modulation (FM) vs
Amplitude Modulation (AM)
FM: all information in the phase or frequency of carrier
Non-linear or rapid improvement in reception quality beyond a minimum
received signal threshold: “capture” effect.
Better noise immunity & resistance to fading
Tradeoff bandwidth (modulation index) for improved SNR: 6dB gain for 2x
bandwidth
Constant envelope signal: efficient (70%) class C power amps ok.
AM: linear dependence on quality & power of rcvd signal

Spectrally efficient but susceptible to noise & fading
Fading improvement using in-band pilot tones & adapt receiver gain to
compensate
Non-constant envelope: Power inefficient (30-40%) Class A or AB power amps
needed: ½ the talk time as FM!

Example Analog: Amplitude Modulation


wbn2007 PT PT I

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

wbn2007 PT PT I

Uploaded by

Copyright:

Available Formats

Point-to-Point Wireless Communication (I):

Digital Basics, Modulation, Detection, Pulse

Based upon slides of Sorour Falahati, A. Goldsmith,

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Assuming both symbols equally likely: uA is chosen if

Log-Likelihood => A simple distance criterion!

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

 Different kinds of digital signal are treated

 Ideal distortion less transmission:

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Band-pass Duality => similar problems occur w/

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

The impulse response of the

The stimulus fed to this filter is the series

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Analog Sampling Pulse amplitude

 Sampling theorem: A bandlimited signal with no spectral

 The sampling rate,

is called Nyquist rate.

 In practice need to sample faster than this because the receiving

Signal peak power

Signal power to average

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

 A uniform linear quantizer is called Pulse Code Modulation

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Process of quantizing noise

 M-ary pulse modulation are used for non-binary symbols

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

• Energy per symbol in binary-PAM: A = 10B

Binary PAM 4-ary PAM

 PCM waveforms category:

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Sampling at rate Encoding each q. value to

Quantizing each sampled Mapping every m = log 2 M data bits to a

 Information (data- or bit-) rate: Rb = 1 / Tb [bits/sec]

Rensselaer Polytechnic Institute Rb = mR Shivkumar Kalyanaraman

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

 Major sources of errors:

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Step 1 – waveform to sample transformation Step 2 – decision making

Demodulate & Sample Detect

For bandpass signals Compensation for

Received waveform Baseband pulse

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

 Find optimum solution for receiver design with the

 First, we focus on designing a receiver which maximizes the

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

 Simplify the model (ideal channel assumption):

Rensselaer Polytechnic Institute Shivkumar Kalyanaraman

which is the time-reversed and delayed version of the conjugate of the

Recall: correlation operation is the projection of the received

 Noise PSD is flat (“white”) => total noise power

Different kinds of digital signal are treated

Ideal distortion less transmission:

Sampling theorem: A bandlimited signal with no spectral

The sampling rate,

In practice need to sample faster than this because the receiving

Signal peak power

Signal power to average

A uniform linear quantizer is called Pulse Code Modulation

M-ary pulse modulation are used for non-binary symbols

PCM waveforms category:

Information (data- or bit-) rate: Rb = 1 / Tb [bits/sec]

Major sources of errors:

Find optimum solution for receiver design with the

First, we focus on designing a receiver which maximizes the

Simplify the model (ideal channel assumption):

Noise PSD is flat (“white”) => total noise power

Why do we need a signal space?

Why are we interested in Euclidean distances between signals?

Properties of inner product:

Norm between two signals:

If all K i = 1 , the signal space is orthonormal.

Example: 1-dimensional orthonornal signal space

These are the in-phase & quadrature-phase dimensions of

Note: two symbols, but only one dimension in BPSK.

Number of matched filters (or correlators) is reduced by 1 compared to

Applying Bayes’ rule gives:

Probability of correct decision for a transmitted symbol

For equally probable symbols:

Note: h refers to the constellation shape/direction

Demodulation: extract baseband message from carrier